Aspect-level sentiment classification is a fine-grained sentiment analysis task, which aims to identify the sentiment polarity (e.g., positive, negative or neutral) of a specific aspect term in a sentence. For example, in “The exterior, unlike the food, is unwelcoming.”, the polarities of aspect terms “exterior” and “food” are negative and positive, respectively. This task has many applications, such as assisting customers to filter online reviews or make purchase decisions on e-commerce websites.
Recent studies have shown that syntactic information such as dependency trees is very effective in capturing long-range syntactic relations that are obscure from the surface form Zhang et al. (2018). Several successful approaches employed graph neural network (GNN) Kipf and Welling (2016) model over dependency trees to aspect-level sentiment classification Huang and Carley (2019); Zhang et al. (2019); Sun et al. (2019); Wang et al. (2020b), which demonstrate that syntactic information is helpful for associating the aspect term with relevant opinion words more directly for increased robustness in sentiment classification.
However, existing approaches are vulnerable to parsing errors Wang et al. (2020b). For example, in Figure 1, the blue parse above the sentence can mislead models to predict negative sentiment for the aspect term “food” with its direct association to “unwelcoming”. Despite their high edge-wise parsing performance on standard benchmarks, state-of-the-art dependency parsers usually struggle to predict flawless parse trees especially in out-of-domain settings. This poses great challenge to dependency-based methods that rely on these parse trees—the added benefit from syntactic structure does not always prevail the noise introduced by model-predicted parses He et al. (2017); Sachan et al. (2020).
In this paper, we propose GraphMerge, a graph ensemble technique to help dependency-based models mitigate the effect of parsing errors. Our technique is based on the observation that different parsers, especially ones with different inductive biases, often err in different ways. For instance, in Figure 1, the green parse under the sentence is incorrect around “unlike the food”, but it nevertheless correctly associates “unwelcoming” with the other aspect term “exterior”, and therefore is less likely to mislead model predictions. Given dependency trees from multiple parses, instead of assigning each dependency tree a separate set of model parameters and ensembling model predictions or dependency-based representations of the same input, we propose to combine the different dependency trees before applying representation learners such as GNNs.
Specifically, we take the union of the edges in all dependency trees from different parsers to construct an ensemble graph, before applying GNNs over it. This exposes the GNN model to various graph hypotheses at once, and allows the model to learn to favor edges that contribute more to the task. To retain the syntactic dependency information between words in the original dependency trees, we also define two different edge types—parent-to-children and children-to-parent—which are encoded by applying relational graph attention networks (RGAT) Busbridge et al. (2019) on the ensemble graph.
Our approach has several advantages. Firstly, since GraphMerge combines dependency trees from different parsers, the GNN models can be exposed to multiple parsing hypotheses and learn to choose edges that are more suitable for the task from data. As a result, the model is less reliant on any specific parser and more robust to parsing errors. Secondly, this improved robustness to parsing errors at no additional computational cost, since we are still applying GNNs to a single graph with the same number of nodes. Last but not least, GraphMerge helps prevent GNNs from overfitting by limiting over-parameterization. Aside from keeping the GNN computation over a single graph to avoid separate parameterization for each parse tree, GraphMerge also introduces more edges in the graph when parses differ, which reduces the diameter of graphs. As a result, fewer layers of GNNs are needed to learn good representations from the graph, alleviating the over-smoothing problem Li et al. (2018b).
To summarize, the main contribution of our work are the following:
We propose a GraphMerge technique to combine dependency parsing trees from different parsers to improve model robustness to parsing errors. The ensemble graph enables the model to learn from noisy graph and select correct edges among nodes at no additional computational cost.
We retain the syntactic dependency information in the original trees by parameterizing parent-to-children and children-to-parent edges separately, which improves the performance of the RGAT model on the ensemble graph.
Our GraphMerge RGAT model outperforms recent state-of-the-art work on three benchmark datasets (Laptop and Restaurant reviews from SemEval 2014 and the ACL 14 Twitter dataset). It also outperforms its single-parse counterparts as well as other ensemble techniques.
2 Related Work
Much recent work on aspect-level sentiment classification has focused on applying attention mechanisms (e.g., co-attention, self attention, and hierarchical attention) to sequence models such recurrent neural networks (RNNs)Tang et al. (2015, 2016); Liu and Zhang (2017); Wang et al. (2018); Fan et al. (2018); Chen et al. (2017); Zheng and Xia (2018); Wang and Lu (2018); Li et al. (2018a, c)
. In a similar vein, pretrained transformer language models such as BERTDevlin et al. (2018) have also been applied to this task, which operates directly on word sequences Song et al. (2019); Xu et al. (2019); Rietzler et al. (2019).
In parallel, researchers have also found syntactic information to be helpful for this task, and incorporated it into aspect-level sentiment classification models in the form of dependency trees Dong et al. (2014); He et al. (2018) as well as constituency trees Nguyen and Shirai (2015). More recently, researchers have developed robust dependency-based models with the help of GNNs that operate either directly on dependency trees Huang and Carley (2019); Zhang et al. (2019); Sun et al. (2019), as well as reshaped dependency trees that center around aspect terms Wang et al. (2020b). While most recent work stack GNNs on top of BERT models, Tang et al. (2020) have also reported gains by jointly learning the two with a mutual biaffine attention mechanism.
Despite the success of these dependency-based models, they are usually vulnerable to parse errors since they rely on a single parser. Furthermore, most prior work that leverage GNNs to encode dependency information treats the dependency tree as an undirected graph, therefore ignores the syntactic relation between words in the sentence.
3 Proposed Model
We are interested in the problem of predicting the sentiment polarity of an aspect term in a given sentence. Specifically, given a sentence of words where
is the aspect term, the goal is to classify the sentiment polarity toward the term as positive, negative, or neutral. Applying GNNs over dependency trees is shown effective to solve this problem, however it is vulnerable to parsing errors. Therefore, we propose a GraphMerge technique to utilize multiple dependency trees to improve robustness to parsing errors. In this section, we will first introduce GraphMerge, our proposed graph ensemble technique, then introduce the GNN model over GraphMerge graph for aspect-level sentiment analysis.
3.1 GraphMerge over Multiple Dependency Trees
To allow graph neural networks to learn dependency-based representations of words while being robust to parse errors that might occur, we introduce GraphMerge, which combines different parses into a single ensemble graph. Specifically, given a sentence and different dependency parses , GraphMerge takes the union of the edges from all parses, and constructs a single graph as follows
where is the shared set of nodes among all graphs111This is true for dependency trees as long as parsers share the same tokenization as input. and is the set of edges in (see the right side of Figure 2 for an example).
As a result, contains all of the (directed) edges from all dependency trees, on top of which we can apply the same GNN models when a single dependency tree is used. Therefore, GraphMerge introduces virtually no computational overhead to existing GNN approaches, compared to traditional ensemble approaches where computational time and/or parameter count scale linearly in . What is more, the resulting graph likely contains more edges from the gold parse which correctly captures the syntactic relation between words in the sentence, allowing the GNN to be robust to parse errors from any specific parser. Finally, since contains more edges between words when parses differ than any single parse and reduces the diameter of the graph, it is also more likely that a shallower GNN model is enough to learn good representations, therefore avoiding over-parameterization and thus overfitting from stacking more GNN layers.
3.2 RGAT over Ensemble Graph
To learn node representations from ensemble graphs, we apply graph attention networks (GAT; Veličković et al., 2018)
. In one layer of GAT, the hidden representation of each node in the graph is computed by attending over its neighbors, with a multi-head self-attention mechanism. The representation for wordat the -th layer of GAT can be obtained as follows
Where is the number of attention heads, is the neighborhood of node in the graph, and the concatenation operation. represents the learnable weights in GAT and denotes activation function. is the attention score between node and node with head .
Edge Types. To apply GAT to ensemble graphs, we first add reciprocal edges for each edge in the dependency tree, and label them with parent-to-children and children-to-parent types, respectively. This allows our model to retain the original syntactic relation between words in the sentence. We also follow previous work to add self loop to each node in the graph, which we differentiate from dependency edges by introducing a third edge type.
We adapt Relational GAT (RGAT) to capture this edge type information. Specifically, we encode the edge type information when computing the attention score between two nodes. We assign each edge type an embedding , incorporate it into attention score computation as follows
where is the representation of the type of the edge connecting nodes and . , and are learnable matrices.
3.3 Sentiment Classification
We extract hidden representations from nodes that correspond to aspect terms in the last RGAT layer, and conduct average pooling to obtain . Then we feed it into a two-layer MLP to calculate the final classification scores :
where and denote learnable weight matrices, and
is the number of sentiment classes. We optimize the model to minimize the standard cross entropy loss function, and apply weight decay to model parameters.
3.4 RGAT Input
The initial word node features for RGAT are obtained from a BERT encoder, with positional information from positional embeddings.
BERT Encoder. We use the pre-trained BERT base model as the encoder to obtain word representations. Specifically, we construct the input as “[CLS] + sentence + [SEP] + term + [SEP]” and feed it into BERT. This allows BERT to learn term-centric representations from the sentence during fine-tuning. To feed the resulting wordpiece-based representations into the word-based RGAT model, we average pool representations of subwords for each word to obtain , the raw input to RGAT.
Positional Encoding. Position information is beneficial for this task, especially when there are multiple aspect terms in one sentence, where it helps to locate opinion words relevant to an aspect term. Although the BERT encoder already takes the word position into consideration, it is dampened after layers of Transformers. Therefore, we explicitly encode the absolute position for each word and add it to the BERT output. Specifically, we add a trainable position embedding matrix to before feeding the resulting representation into RGAT.
|BERT||BERT-SPC Song et al. (2019)||84.46||76.98||78.99||75.03||73.55||72.14|
|AEN-BERT Song et al. (2019)||83.12||73.76||79.93||76.31||74.71||73.13|
|BERT+DT||DGEDT-BERT Tang et al. (2020)||86.3||80.0||79.8||75.6||77.9||75.4|
|BERT+RDT||R-GAT+BERT Wang et al. (2020b)||86.60||81.35||78.21||74.07||76.15||74.88|
DT: Dependency Tree; RDT: Reshaped Dependency Tree.
Data & Processing. We evaluate our model on three datasets: Restaurant and Laptop reviews from SemEval 2014 Task 4 (14Rest and 14Lap)222https://alt.qcri.org/semeval2014/task4/ and ACL 14 Twitter dataset (Twitter) Dong et al. (2014). We remove several examples with “conflict” sentiment polarity labels in the reviews. The statistics of these datasets are listed in Table 1. Following previous work, we report the accuracy and macro F1 scores for sentiment classification.
For dependency-based approaches, we tokenize sentences with Stanford CoreNLP Manning et al. (2014), and then parse them with CoreNLP, Stanza Qi et al. (2020), and the Berkeley neural parser Kitaev and Klein (2018). Since the Berkeley parser returns constituency parses, we further convert it into dependency parses using CoreNLP.
Baselines. We compare our GraphMerge model against published work on these benchmarks, including: BERT-SPC Song et al. (2019) feeds the sentence and term pair into the BERT model and uses the BERT outputs for predictions; AEN-BERT Song et al. (2019) uses BERT as the encoder and employs several attention layers. BERT + Dependency tree based models: DGEDT-BERT Tang et al. (2020) proposes a mutual biaffine module to jointly consider the representations learnt from Transformer and the GNN model over the dependency tree; R-GAT+BERT Wang et al. (2020b) reshapes and prunes the dependency tree to an aspect-oriented tree rooted at the aspect term, and then employs RGAT to encode the new tree for predictions. For fair comparison, we report the results of our GraphMerge model using the same data split (without a development set).
To understand the behavior of different models, we also implement several baseline models. In our experiments, we randomly sample 5% training data as held-out development set for hyper-parameter tuning, use the remaining 95% for training and present results of the average and standard deviation numbers from five runs of random initialization on the test set. We consider these baselines:
BERT-baseline which feeds the sentence-term pair into the BERT-base encoder and then applies a classifier with the representation of the aspect term token.
GAT-baseline with Stanza which employs a vanilla GAT model over single dependency tree obtained from Stanza without differentiating edge types. And the initial node features are the raw output of the BERT encoder.
RGAT over single dependency trees, where we apply RGAT models with parent-to-children and child-to-parent edge types over different dependency trees from the CoreNLP, Stanza, and Berkeley parsers. For a fair comparison to our GraphMerge model, the RGAT input comes from BERT encoder plus position embeddings.
Two ensemble models to take advantage of multiple dependency trees, including a Label-Ensemble model which takes the majority vote from three models each trained on one kind of parses, and a Feature-Ensemble model which applies three sets of RGAT parameters, one for each parse, on top of the BERT encoder with their output features concatenated. These models have more parameters and are more computationally expensive compared to the GraphMerge model when operating on the same parses.
We use PytorchPaszke et al. (2019) to implement our models. The GAT implementation is based on Deep Graph Library Wang et al. (2019). During training, we set the learning rate = , batch size = . We use dev data to select the hidden dimension for GAT/RGAT from , the head number in the multi-head self-attention from , and GAT/RGAT layer from . The 2-layer GAT/RGAT models turn out to be the best based on the dev set. We apply dropout Srivastava et al. (2014) and select the best setting from the dropout rate range = . We set the weight of L2 regularization as
. We train the model up to 5 epochs.333Our code will be released at the time of publication.
4.1 Experimental Results
|BERT-baseline||83.43 0.52||74.94 1.37||77.34 0.90||72.77 1.96||73.47 0.89||72.63 0.82|
|GAT-baseline with Stanza||84.29 0.30||75.75 1.11||77.84 0.27||74.0 0.55||73.82 0.70||72.72 0.42|
|RGAT with Stanza||84.53 0.66||77.29 1.42||77.99 0.62||74.35 0.60||73.99 0.48||72.76 0.33|
|RGAT with Berkeley||84.41 0.86||76.63 1.38||78.09 1.24||73.65 1.76||74.07 0.67||72.65 0.74|
|RGAT with CoreNLP||83.86 0.32||76.03 0.88||78.12 1.02||73.86 1.72||73.96 0.93||72.83 0.95|
|Label-Ensemble||84.68 0.95||77.21 1.54||78.40 1.51||74.36 2.45||74.59 0.46||73.51 0.43|
|Feature-Ensemble||84.64 0.77||77.06 1.45||78.68 0.69||74.80 0.92||74.62 0.76||73.61 0.73|
|GraphMerge||85.16 0.53||77.91 0.87||80.00 0.63||76.50 0.64||74.74 0.93||73.66 0.88|
|GraphMerge||85.16 0.53||77.91 0.87||80.00 0.63||76.50 0.64||74.74 0.93||73.66 0.88|
|- Edge type||84.25 0.59||76.15 1.24||78.65 0.51||74.76 0.71||74.37 1.08||73.25 0.85|
|- Position||84.36 0.36||75.92 1.18||78.37 0.31||74.51 0.48||74.28 1.39||73.34 1.35|
|- (Edge type + Position)||84.16 0.31||75.38 0.69||78.09 0.27||74.29 0.64||73.41 0.63||72.52 0.62|
|Edge Intersection||84.59 0.61||77.06 1.07||78.65 0.94||74.86 1.42||74.68 0.83||73.45 0.73|
We first compare our model to previous work following the evaluation protocol in previous work, and report results in Table 2. As we can see, the GraphMerge model achieves best performances on all three datasets. On the Laptop dataset, the GraphMerge model further outperforms baselines by at least 1.42 accuracy and 2.34 Macro-F1 respectively.
Table 3 shows performance comparisons of the GraphMerge model with other baselines in terms of accuracy and Macro-F1. We observe that:
Syntax information benefits aspect-level sentiment classification. All GAT and RGAT models based on dependency trees outperform BERT-baseline on all three datasets. This demonstrates that leveraging syntax structure information is beneficial to this task.
Ensemble models benefit from multiple parses. The Label-Ensemble, Feature-Ensemble, and GraphMerge models achieve better performance compared to their single dependency tree counterparts. This shows that ensemble models benefit from the presence of different parses and thus less sensitive to parse errors from any single parser.
GraphMerge achieves the best performance overall. Our proposed GraphMerge model not only shows consistent improvements over all single dependency tree models, but also surpasses the other two ensemble models without additional parameters or computational overhead, when compared to the single-tree models.
4.2 Model Analysis
We analyze the proposed GraphMerge model from two perspectives: an ablative analysis of model components and an analysis of the change in the dependency graphs after GraphMerge is applied.
4.2.1 Ablation Study
Model components. We conduct ablation studies of our modeling for edge type and position information in Table 4. We observe that: (1) On three datasets, ablating the edge type degrades the performances. It indicates that the syntactic dependency information in original dependency trees is important. Differentiating edges in the ensemble graph provides more guidance to the model about selecting useful connections among nodes. (2) Removing the position embeddings hurts the performances as well. Although the BERT encoder already incorporates position information at its input, this information is dampened over the layers of Transformers. Emphasizing sequence order again before applying RGAT benefits the task.
Edge Union vs. Edge Intersection. While GraphMerge keeps all edges from different dependency parsing trees for the RGAT model to learn to use, this could also result in too much structural noise and adversely impact performance. We therefore compare GraphMerge to edge intersection, which only retains edges that shared by all individual trees when constructing the ensemble graph, which can be thought of distilling syntactic information that an ensemble parser is confident about. We observe from the last row in Table 4 that edge intersection strategy underperforms GraphMerge on all datasets. We postulate that this is because edge intersection over-prunes edges in the ensemble graph and might introduce more disjoint connected components where parsers disagree, which the RGAT model cannot easily recover from.
4.2.2 Graph Structure Analysis
Effect of GraphMerge on Graph Structure. To better understand the effect of GraphMerge on dependency graphs, we conduct statistical analysis on the test set of 14Lap and 14Rest. Specifically, we are interested in the change in the shortest distance between the aspect term and its opinion words on the dependency graphs. For this analysis, we use the test sets with opinion words labeled by Fan et al. (2019) (see Table 5 for dataset statistics).
We summarize analysis results in Figure 3. We observe that: (1) Compared with single dependency tree, the ensemble graph effectively increases the number of one-hop and two-hops cases, meaning the overall distance between the term and opinion words is shortened on both datasets. (2) Shorter distance between the term and opinion words correlates with better performance. With the ensemble graph, the accuracy of one-hop and two-hops cases beats all single dependency tree models. These observations suggest that the ensemble graph from GraphMerge introduces important connectivity to help alleviate overparameterization from stacking RGAT layers, and that the RGAT model is able to make use of the diversity of edges in the resulting graph to improve classification performance.
Note that although shortening distance correlates with improved results, it does not mean that the closer distance is sufficient for better performance. This is because although the BERT model can be seen as a GAT over a fully-connected graph where a word is reachable for all other context words within one hop Wang et al. (2020a), the BERT-baseline model performs worse than dependency-based models. Therefore, encoding the syntactic structure information in dependency trees is crucial for this task. Our GraphMerge model achieves the best results by shortening the graph distance between the aspect term and opinion words with syntactic information.
Case Study. To gain more insight into the GraphMerge model’s behaviour, we find several examples and visualize their dependency trees from three parsers (Figure 4). Due to the space limit, we only show partial dependency trees that contain essential aspect terms and opinion words. These examples are selected from cases that all single dependency tree RGAT models predict incorrectly, but the GraphMerge model predicts correctly.
We observe that in general, the three parsers do not agree in the neighborhood around the aspect term and opinion words in these sentences. As a result, GraphMerge tends to shorten the distance between the aspect term and the opinion words on the resulting graph. For instance, for all examples in Figure 4, the shortest distances between the aspect term and the opinion words are no more than two in the ensemble graphs, while they vary from 2 to 4 in the original parse trees. This could allow the RGAT model to capture the relation between the words without an exessive amount of layers, thus avoiding overfitting.
On the other hand, we observe that the resulting ensemble graph from GraphMerge is more likely to contain the gold parse for the words in question. For instance, in the first two examples, the gold parse for the words visualized in the figure can be found in the ensemble graph (despite no individual parser predicting it in the first example); the third example also has a higher recall of gold parse edges than each parser despite being difficult to parse. This offers the RGAT model with the correct semantic relationship between these words in more examples during training and evaluation, which is often not accessible with those single parse trees.
|Acc → ARS||Acc → ARS|
|BERT||83.04 → 54.82 (↓28.22)||77.59 → 50.94 (↓26.65)|
|RGAT with Berkeley||84.41 → 56.54 (↓27.87)||78.09 → 51.37 (↓26.72)|
|RGAT with CoreNLP||83.86 → 55.76 (↓28.10)||78.12 → 52.27 (↓25.85)|
|RGAT with Stanza||84.53 → 56.34 (↓28.19)||77.99 → 51.20 (↓26.79)|
|GraphMerge||85.16 → 57.46 (↓27.70)||80.0 → 52.90 (↓27.10)|
Aspect Robustness. To study the aspect robustness of the GraphMerge model, we test our model on the Aspect Robustness Test Set (ARTS) datasets proposed by Xing et al. (2020) (see Table 6 for statistics). The datasets enrich the original 14Lap and 14Rest datasets following three strategies: reverse the sentiment of the aspect term; reverse the sentiment of the non-target terms with originally the same sentiment as target term; generate more non-target aspect terms that have opposite sentiment polarities to the target one. They propose a novel metric, Aspect Robustness Score (ARS), that counts the correct classification of the source example and all its variations generated by the above three strategies as one unit of correctness.
We compare three single dependency tree models with the GraphMerge model in Table 7. We directly evaluate the models trained on the original SemEval datasets on ARTS without further tuning. The results indicate that the GraphMerge model shows better aspect robustness than single dependency tree and BERT models.
We propose a simple yet effective graph-ensemble technique, GraphMerge, to combine multiple dependency trees for aspect-level sentiment analysis. By taking the union of edges from different parsers, GraphMerge allows graph neural model to be robust to parse errors without additional parameters or computational cost. With different edge types to capture the original syntactic dependency in parse trees, our model outperforms previous state-of-the-art models, single-parse models, as well as traditional ensemble models on three aspect-level sentiment classification benchmark datasets.
This work was supported by the National Key R&D Program of China under Grant No.2020AAA108600.
- Relational graph attention networks. arXiv preprint arXiv:1904.05811. Cited by: §1.
Recurrent attention network on memory for aspect sentiment analysis.
Proceedings of the 2017 conference on empirical methods in natural language processing, pp. 452–461. Cited by: §2.
- Bert: pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT 2019, pp. pages 4171–4186. Cited by: §2.
- Adaptive recursive neural network for target-dependent twitter sentiment classification. In Proceedings of the 52nd annual meeting of the association for computational linguistics (volume 2: Short papers), pp. 49–54. Cited by: §2, §4.
- Multi-grained attention network for aspect-level sentiment classification. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp. 3433–3442. Cited by: §2.
- Target-oriented opinion words extraction with target-fused neural sequence labeling. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), Minneapolis, Minnesota, pp. 2509–2518. External Links: Cited by: §4.2.2.
- Deep semantic role labeling: what works and what’s next. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Cited by: §1.
Effective attention modeling for aspect-level sentiment classification. In Proceedings of the 27th International Conference on Computational Linguistics, pp. 1121–1131. Cited by: §2.
- Syntax-aware aspect level sentiment classification with graph attention networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pp. 5472–5480. Cited by: §1, §2.
- Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907. Cited by: §1.
- Constituency parsing with a self-attentive encoder. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Melbourne, Australia. Cited by: §4.
- Hierarchical attention based position-aware network for aspect-level sentiment analysis. In Proceedings of the 22nd Conference on Computational Natural Language Learning, pp. 181–189. Cited by: §2.
Deeper insights into graph convolutional networks for semi-supervised learning. In
Thirty-Second AAAI Conference on Artificial Intelligence, Cited by: §1.
- Transformation networks for target-oriented sentiment classification. arXiv preprint arXiv:1805.01086. Cited by: §2.
- Attention modeling for targeted sentiment. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, pp. 572–577. Cited by: §2.
- The stanford corenlp natural language processing toolkit. In Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations, pp. 55–60. Cited by: §4.
- Phrasernn: phrase recursive neural network for aspect-based sentiment analysis. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pp. 2509–2514. Cited by: §2.
PyTorch: an imperative style, high-performance deep learning library. In Advances in Neural Information Processing Systems 32, H. Wallach, H. Larochelle, A. Beygelzimer, F. dAlché-Buc, E. Fox, and R. Garnett (Eds.), pp. 8024–8035. External Links: Cited by: §4.
- Stanza: a Python natural language processing toolkit for many human languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, External Links: Cited by: §4.
- Adapt or get left behind: domain adaptation through bert language model finetuning for aspect-target sentiment classification. arXiv preprint arXiv:1908.11860. Cited by: §2.
- Do syntax trees help pre-trained transformers extract information?. arXiv preprint arXiv:2008.09084. Cited by: §1.
- Attentional encoder network for targeted sentiment classification. arXiv preprint arXiv:1902.09314. Cited by: §2, Table 2, §4.
Dropout: a simple way to prevent neural networks from overfitting.
The Journal of Machine Learning Research15 (1), pp. 1929–1958. Cited by: §4.
- Aspect-level sentiment analysis via convolution over dependency tree. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pp. 5683–5692. Cited by: §1, §2.
- Effective lstms for target-dependent sentiment classification. arXiv preprint arXiv:1512.01100. Cited by: §2.
- Aspect level sentiment classification with deep memory network. arXiv preprint arXiv:1605.08900. Cited by: §2.
- Dependency graph enhanced dual-transformer structure for aspect-based sentiment classification. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, Online, pp. 6578–6588. External Links: Cited by: §2, Table 2, §4.
- Graph attention networks. In ICLR, Cited by: §3.2.
- Learning latent opinions for aspect-level sentiment classification. In Thirty-Second AAAI Conference on Artificial Intelligence, Cited by: §2.
- Direct multi-hop attention based graph neural network. arXiv preprint arXiv:2009.14332. Cited by: §4.2.2.
- Relational graph attention network for aspect-based sentiment analysis. arXiv preprint arXiv:2004.12362. Cited by: §1, §1, §2, Table 2, §4.
- Deep graph library: a graph-centric, highly-performant package for graph neural networks. arXiv preprint arXiv:1909.01315. Cited by: §4.
- Target-sensitive memory networks for aspect sentiment classification. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 957–967. Cited by: §2.
- Tasty burgers, soggy fries: probing aspect robustness in aspect-based sentiment analysis. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 3594–3605. Cited by: §4.2.2.
- BERT post-training for review reading comprehension and aspect-based sentiment analysis. arXiv preprint arXiv:1904.02232. Cited by: §2.
- Aspect-based sentiment classification with aspect-specific graph convolutional networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pp. 4560–4570. Cited by: §1, §2.
- Graph convolution over pruned dependency trees improves relation extraction. EMNLP. Cited by: §1.
- Left-center-right separated neural network for aspect-based sentiment analysis with rotatory attention. arXiv preprint arXiv:1802.00892. Cited by: §2.